searchkick 4.4.0 → 5.3.1

data/lib/searchkick/reindex_queue.rb

@@ -5,28 +5,53 @@ module Searchkick
     def initialize(name)
       @name = name
 
-      raise Searchkick::Error, "Searchkick.redis not set" unless Searchkick.redis
+      raise Error, "Searchkick.redis not set" unless Searchkick.redis
     end
 
-    def push(record_id)
-      Searchkick.with_redis { |r| r.lpush(redis_key, record_id) }
+    # supports single and multiple ids
+    def push(record_ids)
+      Searchkick.with_redis { |r| r.call("LPUSH", redis_key, record_ids) }
+    end
+
+    def push_records(records)
+      record_ids =
+        records.map do |record|
+          # always pass routing in case record is deleted
+          # before the queue job runs
+          if record.respond_to?(:search_routing)
+            routing = record.search_routing
+          end
+
+          # escape pipe with double pipe
+          value = escape(record.id.to_s)
+          value = "#{value}|#{escape(routing)}" if routing
+          value
+        end
+
+      push(record_ids)
     end
 
     # TODO use reliable queuing
     def reserve(limit: 1000)
-      record_ids = Set.new
-      while record_ids.size < limit && (record_id = Searchkick.with_redis { |r| r.rpop(redis_key) })
-        record_ids << record_id
+      if supports_rpop_with_count?
+        Searchkick.with_redis { |r| r.call("RPOP", redis_key, limit) }.to_a
+      else
+        record_ids = []
+        Searchkick.with_redis do |r|
+          while record_ids.size < limit && (record_id = r.call("RPOP", redis_key))
+            record_ids << record_id
+          end
+        end
+        record_ids
       end
-      record_ids.to_a
     end
 
     def clear
-      Searchkick.with_redis { |r| r.del(redis_key) }
+      Searchkick.with_redis { |r| r.call("DEL", redis_key) }
     end
 
     def length
-      Searchkick.with_redis { |r| r.llen(redis_key) }
+      Searchkick.with_redis { |r| r.call("LLEN", redis_key) }
     end
 
     private
@@ -34,5 +59,22 @@ module Searchkick
     def redis_key
       "searchkick:reindex_queue:#{name}"
     end
+
+    def supports_rpop_with_count?
+      redis_version >= Gem::Version.new("6.2")
+    end
+
+    def redis_version
+      @redis_version ||=
+        Searchkick.with_redis do |r|
+          info = r.call("INFO")
+          matches = /redis_version:(\S+)/.match(info)
+          Gem::Version.new(matches[1])
+        end
+    end
+
+    def escape(value)
+      value.to_s.gsub("|", "||")
+    end
   end
 end

data/lib/searchkick/reindex_v2_job.rb

@@ -1,41 +1,17 @@
 module Searchkick
   class ReindexV2Job < ActiveJob::Base
-    RECORD_NOT_FOUND_CLASSES = [
-      "ActiveRecord::RecordNotFound",
-      "Mongoid::Errors::DocumentNotFound",
-      "NoBrainer::Error::DocumentNotFound",
-      "Cequel::Record::RecordNotFound"
-    ]
-
     queue_as { Searchkick.queue_name }
 
-    def perform(klass, id, method_name = nil, routing: nil)
-      model = klass.constantize
-      record =
-        begin
-          if model.respond_to?(:unscoped)
-            model.unscoped.find(id)
-          else
-            model.find(id)
-          end
-        rescue => e
-          # check by name rather than rescue directly so we don't need
-          # to determine which classes are defined
-          raise e unless RECORD_NOT_FOUND_CLASSES.include?(e.class.name)
-          nil
-        end
-
-      unless record
-        record = model.new
-        record.id = id
-        if routing
-          record.define_singleton_method(:search_routing) do
-            routing
-          end
-        end
-      end
-
-      RecordIndexer.new(record).reindex(method_name, mode: :inline)
+    def perform(class_name, id, method_name = nil, routing: nil, index_name: nil)
+      model = Searchkick.load_model(class_name, allow_child: true)
+      index = model.searchkick_index(name: index_name)
+      # use should_index? to decide whether to index (not default scope)
+      # just like saving inline
+      # could use Searchkick.scope() in future
+      # but keep for now for backwards compatibility
+      model = model.unscoped if model.respond_to?(:unscoped)
+      items = [{id: id, routing: routing}]
+      RecordIndexer.new(index).reindex_items(model, items, method_name: method_name, single: true)
     end
   end
 end

data/lib/searchkick/relation.rb

@@ -0,0 +1,247 @@
+module Searchkick
+  class Relation
+    NO_DEFAULT_VALUE = Object.new
+
+    # note: modifying body directly is not supported
+    # and has no impact on query after being executed
+    # TODO freeze body object?
+    delegate :body, :params, to: :query
+    delegate_missing_to :private_execute
+
+    attr_reader :model
+    alias_method :klass, :model
+
+    def initialize(model, term = "*", **options)
+      @model = model
+      @term = term
+      @options = options
+
+      # generate query to validate options
+      query
+    end
+
+    # same as Active Record
+    def inspect
+      entries = results.first(11).map!(&:inspect)
+      entries[10] = "..." if entries.size == 11
+      "#<#{self.class.name} [#{entries.join(', ')}]>"
+    end
+
+    def execute
+      Searchkick.warn("The execute method is no longer needed")
+      load
+    end
+
+    # experimental
+    def limit(value)
+      clone.limit!(value)
+    end
+
+    # experimental
+    def limit!(value)
+      check_loaded
+      @options[:limit] = value
+      self
+    end
+
+    # experimental
+    def offset(value = NO_DEFAULT_VALUE)
+      # TODO remove in Searchkick 6
+      if value == NO_DEFAULT_VALUE
+        private_execute.offset
+      else
+        clone.offset!(value)
+      end
+    end
+
+    # experimental
+    def offset!(value)
+      check_loaded
+      @options[:offset] = value
+      self
+    end
+
+    # experimental
+    def page(value)
+      clone.page!(value)
+    end
+
+    # experimental
+    def page!(value)
+      check_loaded
+      @options[:page] = value
+      self
+    end
+
+    # experimental
+    def per_page(value = NO_DEFAULT_VALUE)
+      # TODO remove in Searchkick 6
+      if value == NO_DEFAULT_VALUE
+        private_execute.per_page
+      else
+        clone.per_page!(value)
+      end
+    end
+
+    # experimental
+    def per_page!(value)
+      check_loaded
+      @options[:per_page] = value
+      self
+    end
+
+    # experimental
+    def where(value = NO_DEFAULT_VALUE)
+      if value == NO_DEFAULT_VALUE
+        Where.new(self)
+      else
+        clone.where!(value)
+      end
+    end
+
+    # experimental
+    def where!(value)
+      check_loaded
+      if @options[:where]
+        @options[:where] = {_and: [@options[:where], ensure_permitted(value)]}
+      else
+        @options[:where] = ensure_permitted(value)
+      end
+      self
+    end
+
+    # experimental
+    def rewhere(value)
+      clone.rewhere!(value)
+    end
+
+    # experimental
+    def rewhere!(value)
+      check_loaded
+      @options[:where] = ensure_permitted(value)
+      self
+    end
+
+    # experimental
+    def order(*values)
+      clone.order!(*values)
+    end
+
+    # experimental
+    def order!(*values)
+      values = values.first if values.size == 1 && values.first.is_a?(Array)
+      check_loaded
+      (@options[:order] ||= []).concat(values)
+      self
+    end
+
+    # experimental
+    def reorder(*values)
+      clone.reorder!(*values)
+    end
+
+    # experimental
+    def reorder!(*values)
+      check_loaded
+      @options[:order] = values
+      self
+    end
+
+    # experimental
+    def select(*values, &block)
+      if block_given?
+        private_execute.select(*values, &block)
+      else
+        clone.select!(*values)
+      end
+    end
+
+    # experimental
+    def select!(*values)
+      check_loaded
+      (@options[:select] ||= []).concat(values)
+      self
+    end
+
+    # experimental
+    def reselect(*values)
+      clone.reselect!(*values)
+    end
+
+    # experimental
+    def reselect!(*values)
+      check_loaded
+      @options[:select] = values
+      self
+    end
+
+    # experimental
+    def includes(*values)
+      clone.includes!(*values)
+    end
+
+    # experimental
+    def includes!(*values)
+      check_loaded
+      (@options[:includes] ||= []).concat(values)
+      self
+    end
+
+    # experimental
+    def only(*keys)
+      Relation.new(@model, @term, **@options.slice(*keys))
+    end
+
+    # experimental
+    def except(*keys)
+      Relation.new(@model, @term, **@options.except(*keys))
+    end
+
+    # experimental
+    def load
+      private_execute
+      self
+    end
+
+    def loaded?
+      !@execute.nil?
+    end
+
+    def respond_to_missing?(method_name, include_all)
+      Results.new(nil, nil, nil).respond_to?(method_name, include_all) || super
+    end
+
+    # TODO uncomment in 6.0
+    # def to_yaml
+    #   private_execute.to_a.to_yaml
+    # end
+
+    private
+
+    def private_execute
+      @execute ||= query.execute
+    end
+
+    def query
+      @query ||= Query.new(@model, @term, **@options)
+    end
+
+    def check_loaded
+      raise Error, "Relation loaded" if loaded?
+
+      # reset query since options will change
+      @query = nil
+    end
+
+    # provides *very* basic protection from unfiltered parameters
+    # this is not meant to be comprehensive and may be expanded in the future
+    def ensure_permitted(obj)
+      obj.to_h
+    end
+
+    def initialize_copy(other)
+      super
+      @execute = nil
+    end
+  end
+end

data/lib/searchkick/relation_indexer.rb

@@ -0,0 +1,155 @@
+module Searchkick
+  class RelationIndexer
+    attr_reader :index
+
+    def initialize(index)
+      @index = index
+    end
+
+    def reindex(relation, mode:, method_name: nil, full: false, resume: false, scope: nil)
+      # apply scopes
+      if scope
+        relation = relation.send(scope)
+      elsif relation.respond_to?(:search_import)
+        relation = relation.search_import
+      end
+
+      # remove unneeded loading for async and queue
+      if mode == :async || mode == :queue
+        if relation.respond_to?(:primary_key)
+          relation = relation.except(:includes, :preload)
+          unless mode == :queue && relation.klass.method_defined?(:search_routing)
+            relation = relation.except(:select).select(relation.primary_key)
+          end
+        elsif relation.respond_to?(:only)
+          unless mode == :queue && relation.klass.method_defined?(:search_routing)
+            relation = relation.only(:_id)
+          end
+        end
+      end
+
+      if mode == :async && full
+        return full_reindex_async(relation)
+      end
+
+      relation = resume_relation(relation) if resume
+
+      reindex_options = {
+        mode: mode,
+        method_name: method_name,
+        full: full
+      }
+      record_indexer = RecordIndexer.new(index)
+
+      in_batches(relation) do |items|
+        record_indexer.reindex(items, **reindex_options)
+      end
+    end
+
+    def batches_left
+      Searchkick.with_redis { |r| r.call("SCARD", batches_key) }
+    end
+
+    def batch_completed(batch_id)
+      Searchkick.with_redis { |r| r.call("SREM", batches_key, [batch_id]) }
+    end
+
+    private
+
+    def resume_relation(relation)
+      if relation.respond_to?(:primary_key)
+        # use total docs instead of max id since there's not a great way
+        # to get the max _id without scripting since it's a string
+        where = relation.arel_table[relation.primary_key].gt(index.total_docs)
+        relation = relation.where(where)
+      else
+        raise Error, "Resume not supported for Mongoid"
+      end
+    end
+
+    def in_batches(relation)
+      if relation.respond_to?(:find_in_batches)
+        klass = relation.klass
+        # remove order to prevent possible warnings
+        relation.except(:order).find_in_batches(batch_size: batch_size) do |batch|
+          # prevent scope from affecting search_data as well as inline jobs
+          # Active Record runs relation calls in scoping block
+          # https://github.com/rails/rails/blob/main/activerecord/lib/active_record/relation/delegation.rb
+          # note: we could probably just call klass.current_scope = nil
+          # anywhere in reindex method (after initial all call),
+          # but this is more cautious
+          previous_scope = klass.current_scope(true)
+          if previous_scope
+            begin
+              klass.current_scope = nil
+              yield batch
+            ensure
+              klass.current_scope = previous_scope
+            end
+          else
+            yield batch
+          end
+        end
+      else
+        klass = relation.klass
+        each_batch(relation, batch_size: batch_size) do |batch|
+          # prevent scope from affecting search_data as well as inline jobs
+          # note: Model.with_scope doesn't always restore scope, so use custom logic
+          previous_scope = Mongoid::Threaded.current_scope(klass)
+          if previous_scope
+            begin
+              Mongoid::Threaded.set_current_scope(nil, klass)
+              yield batch
+            ensure
+              Mongoid::Threaded.set_current_scope(previous_scope, klass)
+            end
+          else
+            yield batch
+          end
+        end
+      end
+    end
+
+    def each_batch(relation, batch_size:)
+      # https://github.com/karmi/tire/blob/master/lib/tire/model/import.rb
+      # use cursor for Mongoid
+      items = []
+      relation.all.each do |item|
+        items << item
+        if items.length == batch_size
+          yield items
+          items = []
+        end
+      end
+      yield items if items.any?
+    end
+
+    def batch_size
+      @batch_size ||= index.options[:batch_size] || 1000
+    end
+
+    def full_reindex_async(relation)
+      batch_id = 1
+      class_name = relation.searchkick_options[:class_name]
+
+      in_batches(relation) do |items|
+        batch_job(class_name, batch_id, items.map(&:id))
+        batch_id += 1
+      end
+    end
+
+    def batch_job(class_name, batch_id, record_ids)
+      Searchkick.with_redis { |r| r.call("SADD", batches_key, [batch_id]) }
+      Searchkick::BulkReindexJob.perform_later(
+        class_name: class_name,
+        index_name: index.name,
+        batch_id: batch_id,
+        record_ids: record_ids.map { |v| v.instance_of?(Integer) ? v : v.to_s }
+      )
+    end
+
+    def batches_key
+      "searchkick:reindex:#{index.name}:batches"
+    end
+  end
+end